fast_gsw

A ctypes-based Python interface for fast, vectorized execution of a few TEOS-10 GSW (Gibbs SeaWater) oceanographic functions, intended for processing large arrays of CTD (conductivity/temperature/pressure) data, such as from ocean gliders.

Currently implemented:

• rho(C, t, P, lon, lat) — in-situ density
• pot_rho(C, t, P, lon, lat) — potential density (reference pressure 0 dbar)
• SA(C, t, P, lon, lat) — absolute salinity
• CT(C, t, P, lon, lat) — conservative temperature

All inputs/outputs are plain numpy arrays (scalars are broadcast); units are conductivity in mS/cm, temperature in °C, pressure in dbar, and longitude / latitude in decimal degrees.

Why

The pure-Python gsw package computes in-situ density (and similar derived quantities) as a chain of several composed Python function calls per array (e.g. SP_from_C -> SA_from_SP -> rho_t_exact). For large arrays of CTD triplets, fast_gsw collapses that whole pipeline into a single C function call by vendoring and compiling the TEOS-10 C implementation and calling it via ctypes, which removes the Python-level per-step call overhead.

Install

pip install fast_gsw

This compiles the vendored TEOS-10 C sources at install time, so a C compiler (gcc/clang) must be available. No other manual steps, root access, or system-wide library installation are required.

To install from a local checkout:

pip install .

License

fast_gsw itself is licensed under the GPLv3 (see LICENSE). It vendors and compiles source code from the TEOS-10 GSW-C Oceanographic Toolbox, C version 3.06.16, under the license in src/fast_gsw/_csrc/LICENSE.